Original words first
A sharp summary is useful. The quote, timestamp and source are what make it trustworthy.
About MacroThesis
Markets have a long memory. The internet doesn't.
MacroThesis is building a living, sourced record of the people who shape how we understand macro, markets, credit, resources and digital assets.
How it started
MacroThesis started with a practical problem: keeping up with serious macro research meant following the same market thinkers across dozens of podcasts—and relying on memory to connect what they said over time.
One of us followed macro and markets closely. The other built software systems. We kept running into the same gap: valuable views were buried in long conversations, scattered across different shows, and difficult to trace back to the original source.
We built MacroThesis to solve that problem: a sourced, longitudinal record of what market thinkers actually said, organized by topic and supported by verbatim quotes, timestamps, and links to the original conversation.
It is the research tool we wanted to use ourselves.
What we are building
Not one model making a summary. Five agents, each with a job.
Behind MacroThesis is a system of specialized AI agents. One finds every appearance. Another turns speech into structured claims. A verification agent rejects any quote that does not match the transcript word for word. The final stages connect each claim to assets and topics, then compare it with the past to surface key calls, trade ideas and the moment a thesis changes.
- 01Gather
- 02Distill
- 03Verify
- 04Structure
- 05Intelligence
Memory beats snapshots
A market view matters more when you can see what came before, what changed and what stayed consistent.
Tools, not instructions
We organize public ideas for research. We do not tell readers what to buy, sell or believe.
Our role
We report the thinking. You decide what it means.
The views and trade ideas on MacroThesis belong to the people who expressed them. Our job is to preserve the words, attribution and context—not to recommend or endorse a position. Automated processing can make mistakes, and the original source always controls.